Research On Peer-to-peer Network Search Model Based On Gnutella

P2P(Peer-to-Peer) is a new network computing technology. At present, the focus on P2P network research is how to search and locate the resource in the P2P network. In this paper, a new searching model named Dual-Gnutella is designed on the basis of Gnutella model, which achieves efficient resource search ability in the P2P network.A double-layer network structure is used in Dual-Gnutella model, in which all nodes are first classified by their physical location on the Internet. Those nodes whose physical location is close are put into the same class.In each class nodes are divided into super node and common node according to its performance. The super node group serves as the central server and performs resource search in a class using the centralized method. The interconnection between virtual node which is respecting a class forms the superstructure, which performs resource search between classes. The search efficiency in the superstructure is a key issue to the performance of the entire Dual-Gnutella model. Aiming at the weak dynamism of the superstructure,the LightFlood algorithm is improved and the TreeFlood algorithm is given and applied to the resource search in the superstructure. Queries are excuted in two stages using different search strategy to achieve efficient resource search ability in the superstructure.Based on the network simulation software NS2, emulation is performed to simulate the queries with TreeFlood algorithm in superstructure. The simulation results demonstrate that the TreeFlood algorithm has more efficiently reduced the redundant messages compared with the LightFlood algorithm, and the overall search efficiency of Dual-Gnutella model has been considerably improved.